| Metric | cnn | lstm | xgboost | naive |
|---|---|---|---|---|
| diff_of_means | -0.838 | 1.370 | 1.534 | 1.575 |
| ratio_of_sd | 0.980 | 0.969 | 0.909 | 0.914 |
| amplitude_ratio_of_means | 0.816 | 0.779 | 0.609 | 0.641 |
| maximum_error | 0.208 | 0.148 | 0.154 | 0.173 |
| ks_mean_on_coarse_res_with_extremes | 0.196 | 0.179 | 0.347 | 0.323 |
| qqplot_mae | 0.048 | 0.058 | 0.093 | 0.103 |
| acf_mae | 0.087 | 0.094 | 0.140 | 0.126 |
| extremogram_mae | 0.043 | 0.034 | 0.074 | 0.065 |
Important: Right now we are only estimating the upper tail
extremogram. Currently we didn’t find a way to estimate the two tales at
the same time. We are using quant = .97
| Metric | lstm | naive | xgboost | cnn |
|---|---|---|---|---|
| diff_of_means | 0.500 | 0.718 | 0.758 | -1.629 |
| ratio_of_sd | 0.962 | 0.909 | 0.899 | 0.978 |
| amplitude_ratio_of_means | 0.779 | 0.641 | 0.609 | 0.817 |
| maximum_error | 0.176 | 0.174 | 0.152 | 0.213 |
| ks_mean_on_coarse_res_with_extremes | 0.109 | 0.245 | 0.324 | 0.175 |
| qqplot_mae | 0.072 | 0.115 | 0.106 | 0.079 |
| acf_mae | 0.087 | 0.118 | 0.130 | 0.079 |
| extremogram_mae | 0.013 | 0.044 | 0.041 | 0.015 |
Important: Right now we are only estimating the upper tail
extremogram. Currently we didn’t find a way to estimate the two tales at
the same time. We are using quant = .97
| Metric | cnn | lstm | xgboost | naive |
|---|---|---|---|---|
| diff_of_means | -0.885 | 1.373 | 1.649 | 1.685 |
| ratio_of_sd | 0.969 | 0.955 | 0.891 | 0.897 |
| amplitude_ratio_of_means | 0.829 | 0.786 | 0.610 | 0.643 |
| maximum_error | 0.218 | 0.173 | 0.146 | 0.163 |
| ks_mean_on_coarse_res_with_extremes | 0.151 | 0.126 | 0.310 | 0.279 |
| qqplot_mae | 0.062 | 0.068 | 0.107 | 0.118 |
| acf_mae | 0.068 | 0.075 | 0.122 | 0.111 |
| extremogram_mae | 0.037 | 0.041 | 0.071 | 0.066 |
Important: Right now we are only estimating the upper tail
extremogram. Currently we didn’t find a way to estimate the two tales at
the same time. We are using quant = .97
| Metric | xgboost | naive | lstm | cnn |
|---|---|---|---|---|
| diff_of_means | -4.327 | -4.509 | -4.754 | -6.878 |
| ratio_of_sd | 0.971 | 0.990 | 1.029 | 1.055 |
| amplitude_ratio_of_means | 0.638 | 0.665 | 0.804 | 0.848 |
| maximum_error | 0.180 | 0.135 | 0.201 | 0.248 |
| ks_mean_on_coarse_res_with_extremes | 0.405 | 0.388 | 0.274 | 0.263 |
| qqplot_mae | 0.173 | 0.178 | 0.188 | 0.272 |
| acf_mae | 0.132 | 0.128 | 0.094 | 0.095 |
| extremogram_mae | 0.135 | 0.131 | 0.093 | 0.093 |
Important: Right now we are only estimating the upper tail
extremogram. Currently we didn’t find a way to estimate the two tales at
the same time. We are using quant = .97
| Metric | cnn | xgboost | naive | lstm |
|---|---|---|---|---|
| diff_of_means | 32.753 | 36.022 | 36.479 | 38.667 |
| ratio_of_sd | 0.641 | 0.546 | 0.610 | 0.671 |
| amplitude_ratio_of_means | 0.645 | 0.368 | 0.522 | 0.607 |
| maximum_error | 0.166 | 0.269 | 0.239 | 0.196 |
| ks_mean_on_coarse_res_with_extremes | 0.062 | 0.373 | 0.233 | 0.040 |
| qqplot_mae | 1.295 | 1.425 | 1.442 | 1.529 |
| acf_mae | 0.031 | 0.128 | 0.084 | 0.049 |
| extremogram_mae | 0.027 | 0.045 | 0.026 | 0.009 |
Important: Right now we are only estimating the upper tail
extremogram. Currently we didn’t find a way to estimate the two tales at
the same time. We are using quant = .97
| Metric | cnn | xgboost | naive | lstm |
|---|---|---|---|---|
| diff_of_means | 33.020 | 36.673 | 37.285 | 39.359 |
| ratio_of_sd | 0.631 | 0.529 | 0.599 | 0.658 |
| amplitude_ratio_of_means | 0.647 | 0.361 | 0.526 | 0.604 |
| maximum_error | 0.177 | 0.273 | 0.213 | 0.225 |
| ks_mean_on_coarse_res_with_extremes | 0.043 | 0.323 | 0.204 | 0.102 |
| qqplot_mae | 1.305 | 1.451 | 1.474 | 1.556 |
| acf_mae | 0.045 | 0.115 | 0.067 | 0.040 |
| extremogram_mae | 0.030 | 0.032 | 0.019 | 0.028 |
Important: Right now we are only estimating the upper tail
extremogram. Currently we didn’t find a way to estimate the two tales at
the same time. We are using quant = .97
| Metric | cnn | xgboost | naive | lstm |
|---|---|---|---|---|
| diff_of_means | 32.468 | 35.896 | 36.372 | 38.496 |
| ratio_of_sd | 0.646 | 0.543 | 0.607 | 0.674 |
| amplitude_ratio_of_means | 0.657 | 0.371 | 0.525 | 0.624 |
| maximum_error | 0.161 | 0.278 | 0.227 | 0.217 |
| ks_mean_on_coarse_res_with_extremes | 0.074 | 0.357 | 0.221 | 0.089 |
| qqplot_mae | 1.284 | 1.420 | 1.438 | 1.522 |
| acf_mae | 0.043 | 0.125 | 0.084 | 0.047 |
| extremogram_mae | 0.052 | 0.027 | 0.027 | 0.036 |
Important: Right now we are only estimating the upper tail
extremogram. Currently we didn’t find a way to estimate the two tales at
the same time. We are using quant = .97